INTRODUCTION

The demand for microfiltration (MF) and ultrafiltration (UF) in the water and wastewater treatment industry has increased greatly since the mid 1990s, due to the requirement for technology to comply with new disinfection standards for drinking water and more stringent regulations for wastewater discharge, respectively [1]. UF membranes, for example, have been reported to combine high pathogen removal ability with minimal energy consumption [2]. Additionally, application of MF/UF membranes in membrane bioreactors (MBR) has sharply increased in the recent years due to their ability to combine biological treatment and filtration in one compact process. A survey on MBR global market reports that this technology is expected to increase from $216.6 million in 2005 to $488 million in 2013 [3]. MF/UF has also been considered as a promising alternative for the pre-treatment to seawater desalination in the last ten years. MF/UF pre-treatment presents various advantages over the conventional pre-treatment techniques, such as environmental friendship in terms of limiting the sludge amount and use of chemicals, lower foot print and cost effectiveness.

Membrane fouling is the term used to describe the undesirable formation of deposits on membrane surfaces. Various membrane fouling types can be broadly classified as organic fouling, inorganic fouling, particulate deposition (colloidal) and biofouling [4]. Some of the important mechanisms of membrane fouling are, adsorption of feed components, clogging of pores, chemical interaction between solutes and membrane material, gel formation, and bacterial growth [5]. In general, fouling can be controlled by suitable pre-treatment of the feed and membrane cleaning strategies. Over longterm operation, membrane fouling is not completely reversible by simple physical cleaning (i.e. backwash and relaxation) therefore, chemical cleaners are used to remove remaining foulants from the membrane surface and restore the permeate flux. On the other hand, cleaning chemicals can potentially lead to removal or degradation of the membrane material, additives and alteration of membrane properties [6]. Generally, polymeric membranes are naturally hydrophobic and must be blended with additives and pore-formers to increase their hydrophilicity such as polyvinylpyrrolidone (PVP). However, the major concern associated with chemical cleaning remains the potential membrane degradation (or ageing) upon repeated fouling-cleaning cycles.

Although previous reports have reported possible chemical reactions between chemical agents and membrane, little is known on the nascent changes occurring at the molecular and structural level of membrane in progressive membrane ageing over prolonged period. Therefore, this study aimed to:

(1) adapt existing characterisation techniques capable of detecting nascent changes of membrane material to determine the loss of membrane integrity,

(2) carry out an arbitrary ranking of membrane stability against different chemical cleaning agents,

(3) gain a better understanding of the mechanisms occurring during membrane ageing,

(4) identify the most relevant membrane property capable to indicate these changes in membrane structural properties,

(5) assess the relative impact of fouling and cleaning on the filtration behaviour,

(6) assess current membrane cleaning protocols for polymeric microporous membranes during long-term progressive degradation of performance in parallel with membrane ageing and morphological changes and,

(7) characterise and possibly intervene the exact extent of the damage encountered by industrially-aged membranes during their lifetime.

MAIN OBJECTIVES AND OUTCOMES FROM INDIVIDUAL WORK PACKAGES

2.1 Critical assessment of the techniques characterising nascent changes in microporous membrane nature (work package 2 – see Appendix 1)

Various analytical tools are being used to observe and measure the mechanical, morphological and chemical effects of membrane ageing. The desirable characteristics of these methods generally include simplicity and lower cost. It is, however, important to use more sensitive and accurate methods jointly with novel and emerging technologies as it could allow the detection of nascent changes of the membrane. A thorough evaluation of the membrane characterisation techniques was executed by developing a ranking system to compare different criteria for the selection of the most suitable methods to assess membrane degradation or modification, as the result of chemical cleaning. The evaluation of the various analytical methods was reflected by the relative sensitivity of the tools in detecting the changes between new and aged flat sheet and hollow fibre polyvinylidene fluoride (PVDF) membranes. These membranes where statically aged in 2% NaOCl solution for the duration of one week. Clean water tests, contact angle, porometry, tensile test, Thermogravimetric analysis (TGA) and fourier transform infrared spectroscopy (FTIR) were completed on both flat sheet and hollow fibre membranes. Experimental reliability of the membrane characterisation techniques was also rigorously assessed for these membranes. The results from the sensitivity and reliably tests were used to rank all the tested characterisation techniques accordingly. Table 1 shows a summary of the recommended characterisation techniques for membrane ageing.

It was found that the various analytical techniques tested present different degrees of sensitivity and reliability towards the impact of the chemical agents. Therefore, the suitability of different membrane characterisation techniques as predictive tools for membrane ageing depends on membrane configuration (flat sheet/hollow fibre) and nature (material, average pore size). For example, clean water test was found to be more reliable for hollow fibre membranes, while contact angle was more suited for flat sheet membranes. Moreover, generally, different instruments operating range is limited, which makes them inappropriate for some samples. For instance, porometry analysis was only possible for membranes featuring pore size larger than 0.1 µm.

Finally, the relative contradiction observed between the testing methods indicated their limitations in measuring a single parameter. For example, many of the membrane characteristics (such as pore size, charge and roughness) can affect the contact angle value obtained from the goniometer. On the other hand, trends obtained from contact angle, membrane resistance and bubble point tests were successfully used to complement some proposed hypothesis regarding degradation mechanisms.

Findings in this section implied that in order to have a comprehensive record of the critical changes in membrane characteristics, more frequent sampling could be carried out in future work. Ageing tests at lower concentrations could also represent a better assessment of the nascent changes in the membrane properties in the early stages of degradation. Finally, characteristics such as: mean pore diameter, breaking force and ultimate elongation can also be suitable in complementing the results.

2.2 Static ageing for better understanding of degradation mechanisms (work package 3 – see Appendix 2)

From the knowledge gained during the literature review, it was observed that most research studies focused on effects of chemical agents on polystyrene/polyethersulfone (PS/PES), polypropylene (PP) and polyamide (PA) membranes. Very few studies have investigated the effects of cleaning chemicals on PVDF membranes, although this membrane material is becoming more widely used in the water industry. Additionally, limited information is available on the effects of all the main chemicals in the mixture of cleaning solutions in the commercial industry, namely caustics, citric acid and surfactants.

Four different cleaning agents (NaOCl, SDS, HCl and NaOH) were used to age five commercial porous membranes fabricated from various materials. The hollow fibre (HF) membranes used in this study included PVDF-HF, PP and PES. In addition, PVDF-hydrophilic (HPI) and –hydrophobic (HPO) flat sheet membranes were also used in the ageing tests. The maximum duration of the ageing experiment was 30 days, and membrane samples were analysed for a wide range of cleaning intensities. The results obtained from the membrane characterisation allowed (1) an arbitrary ranking of membrane stability against different chemical cleaning agents (2) a better understanding of the mechanisms occurring during membrane ageing. A summary of the assessment of the various membrane samples modification/degradation mechanisms by different cleaning agents is reported in Table 2.

The membrane ranking process indicated that PVDF-HF membrane properties were significantly modified after NaOCl and NaOH; whereas, SDS-aged membranes were not significantly affected. Both PVDF flat sheet membranes (PVDF-HPI and PVDF-HPO) were more sensitive to HCl exposure.

All oxidising agents (NaOCl, HCl and NaOH) generally resulted in the removal of membrane additives. For example, PVDF-HF membrane additives were removed by all three oxidising agents, although the extent of degradation of additives varied for different cleaning chemicals. NaOCl attack generally resulted in loss of membrane integrity through cross linking in PVDF, chain scission in PP, and chlorination of PES.

Cleaning with surfactants (SDS) led to subtle changes in membrane surface characteristics due to attachment of its hydrophobic tail on the membrane surface, which resulted in more hydrophilic and negatively-charged membrane. The membranes exposure to the inorganic acid seemed to only result in modification of membrane surface properties (hydrophilisation) without having a significant impact on membrane additives. Both caustic solutions resulted in degradation of all membranes core material.

2.3 Validity of accelerated ageing approach in terms of predicting membrane degradation in large-scale applications (work packages 4 & 5 – see appendices 3 and 4)

From earlier observations, it was proposed that ageing of the membrane could be partially responsible for the loss of hydraulic performances (i.e. increase in transmembrane pressure (TMP) during constant flux operation) observed during long-term operation [6]. A series of experiments was therefore especially designed to assess for the first time the relative impact of fouling and ageing on the changes of membrane performances. The membrane fouling/cleaning operation was conducted using a cross-flow tubular membrane module and a filtration set up, where all the equipment was pc-controlled. Due to time restrain, the chemical cleaning and rinsing time was artificially long and extensive, compared to the duration of the filtration. Feed solution was used to rinse the cleaning solution from the module after cleaning phase.

Sodium hypochlorite (NaOCl) was used during consecutive fouling and cleaning cyclical experiments. Following these experiments, a range of selected analytical techniques were carried out, allowing a complete assessment of the membrane state after ageing. Benchmark testings were also carried out in order to (1) identify the most relevant membrane property capable to indicate nascent changes in membrane chemical structure and (2) assess the relative impact of fouling and cleaning on the filtration behaviour.

Although the intensity of the chemical attack (up to 510,000 ppm.h) was generally lower than in the previous experiments (see work package 3 and [7]), degradation mechanisms of early stage of ageing was assessed from the current results. The apparent pore size enhancement of the PVDF membrane, obtained from resistance and total organic carbon (TOC) rejection measurements, was due to the combined effect of increase wetting and hydrophilicity of the membrane. Therefore, additives were expected to play a major role in the initial wetting of the membrane, while their presence during consecutive filtration is less crucial to suitable operation. Further characterisation of the various membrane samples indicated that NaOCl exposure lead to the potential damage of the PVDF membrane through cross linking. The fouling material used in these experiments potentially played a protective role for the membrane, as NaOCl was mainly used to oxidise the fouling rather than the membrane material.

Decline in PP membrane performances was significantly more pronounced, when compared to those obtained with the PVDF material. After filtration cleaning cycles, it was observed that the membrane resistance and bubble point pressure decreased significantly, indicating occurrence of pore enhancement. For PP membrane, the ageing mechanism potentially includes chain scission as a result of oxidisation of the polymer due to NaOCl attack. Moreover, the tensile properties of the PP membrane were also severely reduced after ageing.

In addition, the consecutive fouling/cleaning cyclical tests using NaOCl was conducted on two different industrially-aged PP membranes to assess their potential to sustain further ageing in a controlled (but accelerated) environment. By analysing the detailed characteristics of various membranes during accelerated ageing, it was expected that enough relevant information could be available to estimate the exact advancement of membrane ageing of an old membrane and to use those tests as benchmark for ageing assessment.

TGA analysis indicated that the industrially-aged membranes were relatively degraded, possibly through chain session during their long-term industrial use. The extent of the apparent degradation varied between the two samples. The initial assessment of the industrially-aged membranes via FTIR indicated that, even though the membrane additives still remained after the long-term industrial use of in the membrane material, different levels of membranes degradation (chain session) was evident considering the emerging bonds representing membrane degradation products.

After consecutive cyclical ageing, it was observed that both industrially-aged membranes featured similar hydraulic performances and presented a lower operating TMP compared to the new PP membranes, which can be due to the potential modification of these membranes properties during their lifetime. Although final TOC rejection of all aged membranes improved, formation of different types of fouling on the tested membranes was expected to be responsible for the observed variations in the TOC rejection performances obtained between the membranes.

Additionally, the consecutive cyclical ageing resulted in further degradation with observed decreased molecular weight of membrane material through oxidative degradation due to NaOCl attack. The observed embrittlement of these membranes further confirmed the occurrence of the oxidative degradation and chain scission. Indications of pore enhancement due to NaOCl exposure were also clearly observed.

The extent of membrane degradation during consecutive ageing was more advanced than the ones from static ageing tests. For instance, Figure 1 shows that the decreasing trend of membrane resistance (Rm) was more pronounced during the consecutive ageing study (MQ/NaOCl) compared to static ageing. This was probably due to the greater contact area between NaOCl and the membrane since the cleaning solution was briefly filtered through the membrane during the cleaning time. These observations were in agreement with previous study, in which both static and consecutive ageing tests were carried out using PVDF membrane and NaOCl [7]. In the later study, chemical modification of the membrane was observed to occur during the early stages of ageing, and extended contact time did not feature a significant impact on the membrane core material. Finally, the comparison of the statically-, consecutively- and industrially-aged membranes revealed that the effects of cleaning agent nature and concentration were significantly greater than the influence of ageing duration on membrane degradation.

CONCLUSIONS AND RECOMMENDATIONS

In this study, various aspects of membrane degradation were investigated through static and consecutive ageing tests conducted with different types of membrane and cleaning solutions. The results were also compared with industrially-aged membranes. It was postulated that the outcomes of accelerated ageing studies can be used as a tool to predict the loss of integrity of membranes in largescale plants. However, it was observed that the extent of the apparent degradation varied due to samples history. This revealed a number of limitations to be addressed in terms of prediction of the large-scale membrane ageing using accelerated ageing trends; including:

1. The nature of the cleaning agent was observed to influence not only the mechanism of the membrane degradation, but also the severity of membrane integrity loss (work package #3)., since the cleaning chemicals applied in the industry were different to those used in this study, results can, therefore, be only compared to a limited extent.

2. The selection of the maximum concentration and duration that can be applied during accelerated ageing studies can create obstacles to match the cleaning intensities that are generally used in large-scale plants. Most of the previous accelerated ageing studies were based on high concentrations of cleaning agents in relatively short-term lab-scale experiments, with the aim to mimic long-term effect of membrane cleaning on membranes. On the other hand, the use of high concentrations of these chemicals has an instantaneous (sometimes annihilating) effect on some membranes or experimental apparatus. For example, relatively low concentration of NaOH in this study was applied due to instant reaction of PVDF when higher concentrations were used. Given the limited time available to conduct those accelerated ageing experiments, the overall cleaning intensity experienced in the industry can hardly been reached in the labs within a reasonable timeframe.